You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@drill.apache.org by Mohit Anchlia <mo...@gmail.com> on 2014/10/22 02:03:34 UTC

Installation

Is Hadoop required to install drill? The architecture page shows dfs/hbase
and not quite clear if it uses hbase to store metadata?

https://cwiki.apache.org/confluence/display/DRILL/Architectural+Overview

Re: Installation

Posted by Tomer Shiran <ts...@maprtech.com>.
I would also add that Drill uses a file system (local, DFS, ...) to store views. 



> On Oct 21, 2014, at 5:31 PM, Aditya <ad...@gmail.com> wrote:
> 
> On Tue, Oct 21, 2014 at 5:15 PM, Mohit Anchlia <mo...@gmail.com>
> wrote:
> 
>> So what I see in the architectural overview page is one of the data sources
>> to drill? Where does drill store all metadata information?
> 
> ​Drill does not have a centralized metadata store. It discovers schema on
> the fly
> from the data.​ That being said, it can work with data sources which have
> defined
> meta data, for example hive.
> 
> 
>> 
>> When using hadoop with drill is it recommended to install drill on hadoop
>> node to take advantage of data locality?
> 
> Definitely. Drill use​s the locality information to schedule work on
> appropriate
> node.

Re: Installation

Posted by Aditya <ad...@gmail.com>.
On Tue, Oct 21, 2014 at 5:15 PM, Mohit Anchlia <mo...@gmail.com>
wrote:

> So what I see in the architectural overview page is one of the data sources
> to drill? Where does drill store all metadata information?
>

​Drill does not have a centralized metadata store. It discovers schema on
the fly
from the data.​ That being said, it can work with data sources which have
defined
meta data, for example hive.


>
> When using hadoop with drill is it recommended to install drill on hadoop
> node to take advantage of data locality?
>

Definitely. Drill use​s the locality information to schedule work on
appropriate
node.

Re: Installation

Posted by Mohit Anchlia <mo...@gmail.com>.
So what I see in the architectural overview page is one of the data sources
to drill? Where does drill store all metadata information?

When using hadoop with drill is it recommended to install drill on hadoop
node to take advantage of data locality?
On Tue, Oct 21, 2014 at 5:07 PM, Jim Scott <js...@maprtech.com> wrote:

> No, Hadoop is not required. If you look through the work spaces there are
> definitions for the local file system.
> e.g. in linux "root" is mapped to / and is not writable but is readable
>
>
> On Tue, Oct 21, 2014 at 7:03 PM, Mohit Anchlia <mo...@gmail.com>
> wrote:
>
> > Is Hadoop required to install drill? The architecture page shows
> dfs/hbase
> > and not quite clear if it uses hbase to store metadata?
> >
> > https://cwiki.apache.org/confluence/display/DRILL/Architectural+Overview
> >
>
>
>
> --
> *Jim Scott*
> Director, Enterprise Strategy & Architecture
> +1 (347) 746-9281
>
>  <http://www.mapr.com/>
> [image: MapR Technologies] <http://www.mapr.com>
>

Re: Installation

Posted by Jim Scott <js...@maprtech.com>.
No, Hadoop is not required. If you look through the work spaces there are
definitions for the local file system.
e.g. in linux "root" is mapped to / and is not writable but is readable


On Tue, Oct 21, 2014 at 7:03 PM, Mohit Anchlia <mo...@gmail.com>
wrote:

> Is Hadoop required to install drill? The architecture page shows dfs/hbase
> and not quite clear if it uses hbase to store metadata?
>
> https://cwiki.apache.org/confluence/display/DRILL/Architectural+Overview
>



-- 
*Jim Scott*
Director, Enterprise Strategy & Architecture
+1 (347) 746-9281

 <http://www.mapr.com/>
[image: MapR Technologies] <http://www.mapr.com>